A Block Chain Eclipse Attack Detection Based On CNN And Defense Method

As a decentralized distributed ledger technology,blockchain technology has been widely noticed by all sectors due to its trustless ideology and characteristics of data sharing and co-construction.With the continuous expansion of its application scenarios,the security challenges faced by blockchain technology have become increasingly evident.Like most software systems,blockchain systems have security issues at the application,scalability,and protocol levels.Among them,network attacks at the protocol layer are the most difficult potential threat to defend against.As a mainstream attack method on blockchain networks,eclipse attacks have always posed a high threat to blockchain networks,and with the continuous improvement of the underlying network mechanisms of blockchains,attackers may combine eclipse attacks with other attack methods such as routing attacks to attack victim nodes in order to lower the attack threshold of eclipse attacks and quickly form network partitions.Currently,research on the security of blockchain,both domestically and internationally,mainly focuses on transaction privacy protection,data reliability and other aspects,while research on network-level defense technologies is still lacking.The few detection technologies available are also inadequate in dealing with complex network attack environments and do not have complete defense measures.To enhance the blockchain’s resistance to network attacks such as eclipse attacks,this paper proposes a CNN-based method for detecting and adapting to eclipse attacks.This method builds a real eclipse attack environment to obtain attack traffic and perform traffic extraction.Then,through one-hot encoding,data feature expansion,data visualization,and other methods,the ordinary traffic features are processed into data that can be recognized by the CNN model.Experimental results show that the detection method designed in this paper can accurately identify the existence of eclipse attacks through traffic feature recognition.In addition,this method can promptly clean up the damaged routing table once a solar eclipse attack is detected,and then detect and repair the corresponding nodes affected by the routing attack through post-detection.Experimental results have shown that this method can effectively handle various blockchain partition attacks,including solar eclipse attacks,and greatly enhance the network security of blockchain.